A calibration method for a multi-dimensional force sensor and related equipment
Through nonlinear mapping algorithm and differential geometric analysis technology, combined with dynamic analysis processing, the problem of insufficient feature extraction of multi-dimensional force sensors in complex environments is solved, the measurement accuracy and reliability are improved, and the stability and efficiency of the system are improved.
Patent Information
- Application Number
- CN202411328012.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-24
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-09-24
AI Technical Summary
When processing high-dimensional force information and complex geometric environments, the feature extraction of multi-dimensional force sensors is not comprehensive enough and the feature analysis is not in-depth enough, resulting in limited performance optimization, affecting the stable operation and overall efficiency of the system.
A nonlinear mapping algorithm is used for feature extraction, and the feature curvature is adjusted through differential geometric analysis and optimization algorithm, error compensation is performed, and a comprehensive calibration strategy is formulated to match geometric position information to improve calibration results.
It improves the comprehensiveness and in-depthness of feature extraction and analysis of multi-dimensional force sensors in high-dimensional force information and complex geometric environments, improves measurement accuracy and reliability, improves dynamic compensation capabilities, and ensures stable operation and overall efficiency of the system.
Smart Images

Figure CN118857551B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of multi-dimensional force sensors, and in particular to a calibration method and related equipment for a multi-dimensional force sensor. Background Art
[0002] International research on multi-dimensional force sensors has a history of more than 40 years, especially in many countries and regions in the United States, Europe and Asia. Multi-dimensional force sensors have been widely used in industrial automation, robotics, healthcare and motion analysis due to their ability to accurately capture and quantify forces in different directions.
[0003] Among the relevant technical means, the use of calibration robots to perform standardized testing and calibration of multi-dimensional force sensors is currently an important method to improve the accuracy and reliability of sensors. The calibration robot combines a preset calibration algorithm with a precision robotic arm to automatically test and adjust the sensor output to simulate the force conditions in the actual working environment. This technology not only improves the calibration efficiency and repeatability, but also reduces human errors through the automation process, providing a reliable guarantee for the high-precision use of sensors.
[0004] Regarding the above technical solution, although automated calibration can effectively improve the calibration speed and accuracy, when processing high-dimensional information and complex geometric environments, there are still problems such as incomplete feature extraction, insufficient depth of feature analysis, and limited dynamic compensation capabilities, which in turn limits the performance optimization of the sensor in actual applications, affecting the stable operation and overall efficiency of the system. Summary of the invention
[0005] In order to improve the problem of insufficiently comprehensive feature extraction and insufficiently in-depth feature analysis when processing high-dimensional force information and complex geometric environments, the present application provides a calibration method and related equipment for a multi-dimensional force sensor.
[0006] The present invention provides a calibration method for a multi-dimensional force sensor, comprising: acquiring force data and geometric position information of the multi-dimensional force sensor; performing feature extraction on the force data using a preset nonlinear mapping algorithm to obtain a comprehensive feature vector; performing differential geometry analysis on the comprehensive feature vector to obtain a characteristic curvature and a characteristic torsion, adjusting the characteristic curvature based on a preset optimization algorithm to obtain an adjusted characteristic curvature and an adjustment parameter; performing error compensation on the characteristic torsion based on the adjusted characteristic curvature and the adjustment parameter to obtain a compensated characteristic torsion and an error compensation parameter; performing dynamic analysis processing on the compensated characteristic torsion and the error compensation parameter to obtain an analysis result; processing the analysis result according to the comprehensive feature vector to obtain a calibration index, and formulating a comprehensive calibration strategy based on the calibration index and the adjustment parameter; performing matching analysis on the geometric position information and the comprehensive calibration strategy to obtain a calibration result.
[0007] As a preferred solution, the step of obtaining the force data and geometric position information of the multi-dimensional force sensor includes: obtaining real-time force data of the multi-dimensional force sensor based on a data module; positioning the multi-dimensional force sensor based on optical positioning technology, calibrating the geometric sensitive area of the multi-dimensional force sensor, and obtaining the morphological parameters and spatial response characteristics of the multi-dimensional force sensor; integrating the morphological parameters and the spatial response characteristics to obtain pre-processed data, and calculating the pre-processed data through a Kalman filter algorithm and a particle filter algorithm to obtain geometric position information.
[0008] As a preferred scheme, the step of extracting features from the force data using a preset nonlinear mapping algorithm to obtain a comprehensive feature vector includes: taking the geometric position information as a geometric position constraint, and adjusting the preset nonlinear mapping algorithm based on the geometric position constraint, inputting the force data into the adjusted nonlinear mapping algorithm, and performing dimensionality reduction analysis on the force data in the adjusted nonlinear mapping algorithm using kernel principal component analysis to obtain a dimensionality reduction result; extracting features from the dimensionality reduction result using a support vector machine to obtain a fused force feature and a force feature correlation; performing feature association analysis on the fused force feature and the force feature correlation to obtain an associated feature vector; performing position-sensitive weighting on the associated feature vector through the geometric position constraint to obtain a position-weighted force feature vector; inputting the position-weighted force feature vector into a preset feature mapping network to obtain a low-dimensional force feature vector reflecting the force characteristics; and fusing the low-dimensional force feature vector with the geometric position constraint to generate a comprehensive feature vector representing multi-dimensional force perception capability.
[0009] As a preferred embodiment, the steps of performing differential geometry analysis on the comprehensive feature vector to obtain characteristic curvature and characteristic torsion, adjusting the characteristic curvature based on a preset optimization algorithm to obtain the adjusted characteristic curvature and adjustment parameters include: inputting the comprehensive feature vector into a preset differential geometry analysis model, performing local analysis on the comprehensive feature vector through surface theory in the differential geometry analysis model, and extracting the characteristic curvature; wherein the characteristic curvature reflects the degree of bending of the multi-dimensional force sensor under different forces; calculating the characteristic torsion by using torsion theory and combining the characteristic curvature; wherein the characteristic torsion reveals the intrinsic properties of the geometric deformation of the force sensor; determining the optimal characteristic curvature of the characteristic torsion to characterize the geometric deformation of the sensor through an optimization algorithm, taking the optimal characteristic curvature as the adjusted characteristic curvature, and determining and recording the adjustment parameters of the optimal characteristic curvature of the characteristic torsion to characterize the geometric deformation of the sensor.
[0010] As a preferred scheme, the step of performing error compensation on the characteristic torsion based on the adjusted characteristic curvature and adjustment parameters to obtain compensated characteristic torsion and error compensation parameters includes: inputting the adjusted characteristic curvature and the adjustment parameters into an error compensation model, analyzing the adjusted characteristic curvature and the adjustment parameters in the error compensation model using a back propagation network to obtain result data; extracting the error source affecting the accuracy of the characteristic torsion from the result data to obtain error information; wherein the error source includes any one of the following: system error, environmental error, measurement error; and correcting the characteristic torsion and adjustment parameters based on the error information to obtain compensated characteristic torsion and error compensation parameters.
[0011] As a preferred solution, the steps of dynamically analyzing and processing the compensated characteristic torsion and the error compensation parameters to obtain analysis results; processing the analysis results according to the comprehensive characteristic vector to obtain calibration indicators, and formulating a comprehensive calibration strategy based on the calibration indicators and the adjustment parameters include: inputting the compensated characteristic torsion and the error compensation parameters into a preset dynamic analysis model, executing a time series algorithm in the dynamic analysis model to evaluate the time dependence of the compensated characteristic torsion, capturing the pattern of sensor performance changes over time, and obtaining a timing analysis result; performing a stability analysis on the timing analysis results based on a preset stability analysis algorithm to confirm the consistency and predictability of the compensated characteristic torsion under different operating conditions, and obtaining a stability analysis result; integrating the timing analysis results and the stability analysis results by fusing a real-time feedback algorithm to obtain a performance analysis result that characterizes the dynamic performance of the sensor; evaluating the performance analysis results according to the comprehensive characteristic vector using a difference evaluation tool, establishing a dynamic response law, and formulating calibration indicators based on the dynamic response law and preset market feedback data; formulating a comprehensive calibration strategy for the multi-dimensional force sensor by combining the calibration indicators with the error compensation parameters.
[0012] As a preferred scheme, the step of matching and analyzing the geometric position information and the comprehensive calibration strategy to obtain a calibration result includes: using a fuzzy logic algorithm to analyze the geometric position information to obtain a spatial positioning error; adjusting the comprehensive calibration strategy according to a preset piecewise linear correction algorithm to obtain calibration parameters; inputting the spatial positioning error and the calibration parameters into a preset intelligent compensation model, and matching the geometric position information and the calibration parameters in the intelligent compensation model using a genetic optimization algorithm to obtain matching analysis data and calibration instructions; performing a real-time calibration operation according to the calibration instructions to obtain real-time calibration data, and performing terminal calibration analysis on the real-time calibration data and the matching analysis data to obtain a calibration result.
[0013] The present application also provides a calibration device for a multi-dimensional force sensor, comprising: an acquisition module, used to acquire force data and geometric position information of the multi-dimensional force sensor; an extraction module, used to perform feature extraction on the force data using a preset nonlinear mapping algorithm to obtain a comprehensive feature vector; an adjustment module, used to perform differential geometry analysis on the comprehensive feature vector to obtain a characteristic curvature and a characteristic torsion, and adjust the characteristic curvature based on an optimization algorithm to obtain an adjusted characteristic curvature and an adjustment parameter; a compensation module, used to perform error compensation on the characteristic torsion based on the adjusted characteristic curvature and the adjustment parameter to obtain a compensated characteristic torsion and an error compensation parameter; an analysis module, used to perform dynamic analysis and processing on the compensated characteristic torsion and the error compensation parameter to obtain an analysis result; processing the analysis result according to the comprehensive feature vector to obtain a calibration index, and formulating a comprehensive calibration strategy based on the calibration index and the adjustment parameter; a matching module, used to perform matching analysis on the geometric position information and the comprehensive calibration strategy to obtain a calibration result.
[0014] The present application also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, the calibration method of the multi-dimensional force sensor described above is implemented.
[0015] The present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the processor executes the calibration method for the multi-dimensional force sensor as described above.
[0016] Compared with the prior art, the present application has the following beneficial effects: comprehensive feature analysis. The force data and geometric position information are obtained through high-precision measuring equipment to ensure the reliability and accuracy of the data, and the nonlinear mapping algorithm is applied to extract the features of the data to obtain a comprehensive feature vector, and the characteristic curvature is adjusted through differential geometry analysis and optimization algorithm to further improve the accuracy of feature extraction; based on the adjusted characteristic curvature and adjustment parameters, the characteristic torsion is error compensated to obtain the compensated high-precision characteristic parameters; through dynamic analysis and processing, combined with the comprehensive feature vector, detailed analysis results and calibration indicators are obtained, and a comprehensive calibration strategy is formulated based on these indicators; the geometric position information is matched and analyzed with the calibration strategy to obtain the final calibration result, so as to improve the measurement accuracy and reliability of the multi-dimensional force sensor, and improve the problem of insufficient feature extraction and insufficient in-depth feature analysis when processing high-dimensional force information and complex geometric environments. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0018] The structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with this technology. They are not used to limit the conditions under which the present invention can be implemented, and therefore have no substantive technical significance. Any structural modification, change in proportion or adjustment of size, without affecting the effects and purposes that can be achieved by the present invention, should still fall within the scope of the technical contents disclosed by the present invention.
[0019] Figure 1 is a flow chart of a calibration method for a multi-dimensional force sensor provided by an embodiment of the present invention;
[0020] Figure 2 is a schematic block diagram of the structure of a calibration device for a multi-dimensional force sensor provided in an embodiment of the present invention;
[0021] Figure 3 It is a schematic block diagram of the structure of an electronic device provided by an embodiment of the present invention.
[0022] Description of reference numerals:
[0023] 10. Calibration device of multi-dimensional force sensor; 11. Acquisition module; 12. Extraction module; 13. Adjustment module; 14. Compensation module; 15. Analysis module; 16. Matching module; 20. Electronic device; 21. Memory; 22. Processor. DETAILED DESCRIPTION
[0024] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0025] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.
[0026] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.
[0027] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.
[0028] The technical solution of the present invention is further described below with reference to the accompanying drawings and through specific implementation methods.
[0029] Embodiment 1:
[0030] like Figure 1 As shown, a calibration method for a multi-dimensional force sensor provided in an embodiment of the present application includes steps S100 to S600.
[0031] Step S100: Acquire force data and geometric position information of the multi-dimensional force sensor.
[0032] In this step, the sensor first needs to be placed in a predetermined calibration environment. Through a series of precise robotic arm operations, the force sensor collects data in different positions and directions. Specifically, the sensor records the force data and corresponding geometric position information of each position point in a preset coordinate system to ensure that the force information under different conditions is accurately captured.
[0033] For example, in a six-degree-of-freedom robotic arm system, the sensor is moved step by step to preset calibration points, a known force is applied to each calibration point, the sensor feedback value is recorded, and the data is stored in a data storage device for subsequent analysis and processing.
[0034] Step S200: extracting features from the force data using a preset nonlinear mapping algorithm to obtain a comprehensive feature vector.
[0035] In this step, the collected force data is processed by using a nonlinear mapping algorithm to extract the key features contained in the data; specifically, the input multi-dimensional force data is trained and mapped using algorithms such as a multi-layer neural network or a support vector machine, and ultimately a comprehensive feature vector is obtained, which represents the main features of the force data.
[0036] For example, a three-layer neural network is used to process force data. The input layer receives the original force data, performs nonlinear transformation through the activation function of the hidden layer, and the output layer generates a comprehensive feature vector that contains the core information of the force data, such as direction, size, etc.
[0037] Step S300: Perform differential geometry analysis on the comprehensive characteristic vector to obtain characteristic curvature and characteristic torsion, and adjust the characteristic curvature based on a preset optimization algorithm to obtain an adjusted characteristic curvature and adjustment parameters.
[0038] In this step, the method of differential geometry is used to conduct an in-depth analysis of the comprehensive eigenvector to obtain more detailed geometric features; specifically, the curvature and torsion formulas in differential geometry theory are used to calculate the eigenvector to obtain its curvature and torsion parameters at different positions, and then these parameters are adjusted through the optimization algorithm to reduce the error.
[0039] For example, the Gaussian curvature and mean curvature formulas in differential geometry are used to calculate the curvature of the comprehensive eigenvector at each calibration point, and then a genetic algorithm is used to optimize and adjust it, ultimately obtaining the optimal curvature parameters and corresponding adjustment parameters.
[0040] Step S400: performing error compensation on the characteristic torsion based on the adjusted characteristic curvature and the adjustment parameter to obtain the compensated characteristic torsion and the error compensation parameter.
[0041] In this step, the adjustment parameters obtained in the previous step are used to perform error compensation on the characteristic torsion to improve the accuracy of the sensor; specifically, by applying the adjusted characteristic curvature to the calculation of the torsion, the error caused by the model deviation is reduced, and the compensated torsion value and the corresponding error compensation parameters are obtained.
[0042] For example, assuming that at a certain calibration point, the initially calculated characteristic torsion rate is 0.05, and after compensating it by adjusting the parameters, the new characteristic torsion rate is 0.045, and the corresponding error compensation parameter is -0.005.
[0043] Step S500, dynamically analyze and process the compensated characteristic torsion and error compensation parameters to obtain analysis results; process the analysis results according to the comprehensive characteristic vector to obtain calibration indicators, and formulate a comprehensive calibration strategy based on the calibration indicators and adjustment parameters.
[0044] In this step, the compensated data is further processed by dynamic analysis methods to obtain more accurate calibration results. Specifically, the compensated characteristic torsion and error parameters are analyzed using time series analysis or dynamic system modeling methods to obtain their changing patterns at different time points. Then, the final calibration index is calculated in combination with the comprehensive characteristic vector.
[0045] For example, the Kalman filter algorithm is used to process the compensated data in real time to obtain the optimal estimate value at each time point. Then, based on these estimates and the comprehensive feature vector, the optimal calibration strategy is formulated to ensure that the sensor can maintain high accuracy in different environments.
[0046] Step S600: Perform matching analysis on the geometric position information and the comprehensive calibration strategy to obtain a calibration result.
[0047] In this step, the calibration strategy obtained previously is applied to the geometric position information of the sensor, and a matching analysis is performed to obtain the final calibration result. Specifically, the effectiveness of the calibration strategy is verified by adjusting and matching the geometric position information, and the sensor is finally calibrated and adjusted based on the matching result.
[0048] For example, in practical applications, the sensor is placed in a complex geometric environment. By applying the calibration strategy to the sensor's position information, a matching analysis is performed to verify its accuracy at different positions. Based on the analysis results, the sensor's calibration is finally adjusted to ensure that it maintains high accuracy under various working conditions.
[0049] In this embodiment, first, the force data and geometric position information of the multi-dimensional force sensor are obtained; then, the force data is feature extracted using a preset nonlinear mapping algorithm to obtain a comprehensive feature vector; then, the comprehensive feature vector is subjected to differential geometry analysis to obtain a characteristic curvature and a characteristic torsion, and the characteristic curvature is adjusted based on a preset optimization algorithm to obtain an adjusted characteristic curvature and an adjustment parameter; then, the characteristic torsion is error compensated based on the adjusted characteristic curvature and the adjustment parameter to obtain a compensated characteristic torsion and an error compensation parameter; next, the compensated characteristic torsion and the error compensation parameter are dynamically analyzed and processed to obtain an analysis result; finally, the analysis result is processed according to the comprehensive feature vector to obtain a calibration index, and a comprehensive calibration strategy is formulated based on the calibration index and the adjustment parameter, and a matching analysis is performed on the geometric position information and the comprehensive calibration strategy to obtain a calibration result.
[0050] By adopting nonlinear mapping algorithms and differential geometry analysis techniques, the comprehensive feature vectors of multi-dimensional force sensors can be effectively extracted, and curvature and torsion adjustments and error compensation can be performed, thereby improving the comprehensiveness and depth of feature extraction and feature analysis of sensors in processing high-dimensional force information and complex geometric environments; dynamic analysis processing and error compensation further enhance the dynamic compensation capability of sensors in practical applications, ensure the stable operation of the system and optimization of overall efficiency, thereby improving the accuracy and reliability of calibration, and improving the problems of insufficiently comprehensive feature extraction and insufficiently in-depth feature analysis when processing high-dimensional force information and complex geometric environments.
[0051] Embodiment 2:
[0052] In step S100, real-time force data of the multi-dimensional force sensor is acquired based on the data module.
[0053] The data module is used to collect the force data of the multi-dimensional force sensor during the working process in real time; specifically, the data module is composed of a high-precision sensor array and a high-speed data acquisition unit. These sensor arrays are installed at different positions of the sensor to obtain multi-dimensional force information in different directions and positions, and the data are transmitted to the central processing unit in real time for processing through the high-speed data acquisition unit.
[0054] For example, the data module may include multiple force sensors mounted on the robotic arm, which can record the forces applied at different positions and directions in real time when the robotic arm performs tasks, and transmit the data to a computer system via wireless or wired communication for subsequent processing and analysis.
[0055] The multi-dimensional force sensor is positioned based on optical positioning technology, and the geometric sensitive area of the multi-dimensional force sensor is calibrated to obtain the morphological parameters and spatial response characteristics of the multi-dimensional force sensor.
[0056] The sensor is spatially positioned by using a high-precision optical positioning system, such as a laser scanner or an infrared positioning system. Specifically, the transmitter and receiver of the optical positioning system are installed in a calibration environment, and the precise position and posture of the sensor are determined by emitting a light beam and receiving a reflected signal. The geometric sensitive area of the sensor is then calibrated in detail to obtain its morphological parameters and spatial response characteristics.
[0057] For example, in a laboratory environment, a laser scanner can be used to scan a multi-dimensional force sensor placed on a workbench to obtain its three-dimensional morphological data and response characteristics at different angles, and store these data in a computer for subsequent analysis and processing.
[0058] The morphological parameters and spatial response characteristics are integrated to obtain preprocessed data, which are then calculated using the Kalman filter algorithm and the particle filter algorithm to obtain geometric position information.
[0059] By integrating morphological parameters and spatial response characteristics into a unified data framework, the Kalman filter algorithm and particle filter algorithm are used to fuse and optimize the data; specifically, the Kalman filter algorithm is used to process continuous-time data to reduce the impact of noise, while the particle filter algorithm is used to process nonlinear and non-Gaussian systems to enhance the robustness of the data, thereby obtaining more accurate geometric position information.
[0060] For example, during the calibration process, the morphological parameters and spatial response characteristics are first integrated into a multidimensional data set, and then the data is preliminarily processed by the Kalman filter algorithm to remove noise and errors. The processed data is then further optimized using the particle filter algorithm to ultimately obtain the precise geometric position information of the sensor.
[0061] In step S200, the geometric position information is used as the geometric position constraint, and the preset nonlinear mapping algorithm is adjusted based on the geometric position constraint. The force data is input into the adjusted nonlinear mapping algorithm, and the kernel principal component analysis is used in the adjusted nonlinear mapping algorithm to perform dimensionality reduction analysis on the force data to obtain a dimensionality reduction result.
[0062] By using geometric position information as a constraint condition, the preset nonlinear mapping algorithm is adjusted to make it more adaptable to the actual geometric environment; specifically, the geometric position information is input into the constraint module of the nonlinear mapping algorithm, and the parameters of the mapping algorithm are adjusted so that it can process the force data more accurately. Then, the kernel principal component analysis method is applied to the adjusted mapping algorithm to reduce the dimensionality of the high-dimensional force data and extract the main characteristic components.
[0063] For example, during the calibration process, the geometric position information of the sensor can be input into an adjusted neural network model. By adjusting the weights and bias parameters of the network, it can better fit the actual data. Then, the kernel principal component analysis method is used to reduce the dimensionality of the input high-dimensional force data to obtain information containing the main features.
[0064] Support vector machine is used to extract features from the dimension reduction results to obtain fusion force features and force feature correlation.
[0065] Further feature extraction is performed by inputting the reduced-dimensional data into a support vector machine model; specifically, the support vector machine extracts features from the reduced-dimensional force data by finding the optimal classification boundary in the data, thereby obtaining a comprehensive feature that integrates force features and force feature correlations.
[0066] For example, in data processing, the force data after dimensionality reduction can be input into a trained support vector machine model. Through the classification and regression capabilities of the model, the main force features and their correlation information in the data can be extracted to form a fused feature vector.
[0067] The fusion force feature and the force feature correlation are analyzed to obtain the correlation feature vector.
[0068] By conducting an in-depth analysis of the fusion force characteristics and the correlation of force characteristics, the intrinsic relationship between them is revealed; specifically, feature association analysis methods, such as correlation analysis and principal component analysis, are used to analyze the fusion force characteristics and their correlation, and the correlation feature vectors reflecting their intrinsic relationship are extracted.
[0069] For example, the principal component analysis method can be used to process the fusion force features and correlation data, extract the main correlation components, and form a correlation feature vector to represent the intrinsic relationship and mutual influence between these features.
[0070] The associated eigenvectors are position-sensitively weighted through geometric position constraints to obtain the position-weighted force eigenvectors.
[0071] By utilizing the geometric position information, the associated eigenvector is weighted to make it more sensitive to the specific geometric position; specifically, the geometric position information is used as a weight factor and applied to each element of the associated eigenvector to make its weight at a specific position higher, thereby obtaining a position-weighted force eigenvector.
[0072] For example, during the calibration process, the geometric position information of the sensor can be used as a weight matrix and multiplied by each element of the associated eigenvector to obtain a new weighted eigenvector whose force characteristics at a specific position are more obvious and accurate.
[0073] The position-weighted force feature vector is input into a preset feature mapping network to obtain a low-dimensional force feature vector reflecting the force characteristics.
[0074] By inputting the position-weighted force feature vector into a preset feature mapping network, further feature mapping and dimensionality reduction processing are performed; specifically, the feature mapping network can be a deep learning model, such as a convolutional neural network or an autoencoder, which extracts a low-dimensional feature vector reflecting the force characteristics through nonlinear transformation of the input feature vector.
[0075] For example, a convolutional neural network model can be used to input the position-weighted force feature vector into the network. After multiple processing by convolutional layers and pooling layers, a low-dimensional force feature vector is finally obtained, which can accurately reflect the force characteristics of the sensor.
[0076] The low-dimensional force feature vector is integrated with the geometric position constraint to generate a comprehensive feature vector representing the multi-dimensional force perception ability.
[0077] By fusing the low-dimensional force feature vector with the geometric position information, a comprehensive feature vector is generated to comprehensively characterize the multi-dimensional force perception capability of the sensor. Specifically, the low-dimensional force feature vector and the geometric position information are concatenated or weighted to form a new comprehensive feature vector that can simultaneously reflect the force characteristics and position characteristics.
[0078] For example, the low-dimensional force feature vector can be concatenated with the geometric position information to form a high-dimensional comprehensive feature vector, which contains the response characteristics of the sensor at different positions and forces, and comprehensively characterizes its multi-dimensional force perception capabilities.
[0079] In step S300, the comprehensive feature vector is input into a preset differential geometry analysis model, and the comprehensive feature vector is locally analyzed by surface theory in the differential geometry analysis model to extract the characteristic curvature; wherein the characteristic curvature reflects the degree of bending of the multi-dimensional force sensor under different forces.
[0080] By utilizing the preset differential geometry analysis model, the comprehensive feature vector is deeply analyzed; specifically, the Gaussian curvature and mean curvature methods in surface theory are used to perform local analysis on the comprehensive feature vector and extract the characteristic curvature of each point, reflecting the degree of bending of the sensor under different forces.
[0081] For example, the Gaussian curvature formula can be used to calculate each element in the comprehensive feature vector to obtain its curvature value at different positions and forces. These curvature values can intuitively reflect the bending degree and deformation characteristics of the sensor.
[0082] The characteristic torsion is calculated by using torsion theory and combining with characteristic curvature. The characteristic torsion reveals the intrinsic properties of the geometric deformation of the force sensor.
[0083] By combining the characteristic curvature and torsion theory, the comprehensive characteristic vector is further analyzed; specifically, the characteristic torsion of each point is calculated using the torsion formula, revealing the geometric deformation characteristics of the force sensor under different forces.
[0084] For example, the torsion formula can be used to calculate each point in the comprehensive feature vector to obtain its torsion value at different positions and forces. These torsion values can reflect the intrinsic deformation characteristics of the sensor.
[0085] The optimal characteristic curvature of the characteristic torsion characterizing the geometric deformation of the sensor is determined by an optimization algorithm, the optimal characteristic curvature is used as the adjusted characteristic curvature, and the adjustment parameters of the optimal characteristic curvature of the characteristic torsion characterizing the geometric deformation of the sensor are determined and recorded.
[0086] The characteristic torsion is optimized by using an optimization algorithm to determine the best characteristic curvature that characterizes the geometric deformation of the sensor; specifically, a genetic algorithm or a particle swarm optimization algorithm is used to optimize the characteristic torsion, and the best characteristic curvature that minimizes the error is found and used as the adjusted characteristic curvature, and the corresponding adjustment parameters are recorded at the same time.
[0087] For example, a genetic algorithm can be used to perform multiple iterations of optimization on the characteristic torsion to find a characteristic curvature value that minimizes the error and use it as the adjusted characteristic curvature. The corresponding genetic algorithm parameters are recorded for subsequent compensation and analysis.
[0088] In step S400, the adjusted characteristic curvature and adjustment parameters are input into the error compensation model, and the adjusted characteristic curvature and adjustment parameters are analyzed by a back propagation network in the error compensation model to obtain result data.
[0089] Error analysis and compensation are performed by inputting the adjusted characteristic curvature and adjustment parameters into a preset error compensation model; specifically, the error compensation model can be a back propagation neural network, which analyzes the adjusted characteristic curvature and adjustment parameters through back propagation calculation of the input data to obtain error compensation result data.
[0090] For example, a back-propagation neural network model can be used to input the adjusted characteristic curvature and adjustment parameters into the network, and through multi-layer back-propagation calculations, the result data after error compensation can be obtained. These data can reflect the source of the error and the compensation effect.
[0091] The error sources that affect the accuracy of the characteristic torsion are extracted from the result data to obtain error information; wherein the error sources include any one of the following: system error, environmental error, and measurement error.
[0092] By analyzing the result data, the error sources that affect the accuracy of the characteristic torsion are extracted; specifically, by analyzing the error components in the result data, different error sources such as system error, environmental error and measurement error are identified to form error information.
[0093] For example, the result data can be statistically analyzed to identify the proportion of systematic errors, the impact of environmental errors, and the fluctuation range of measurement errors, and obtain a detailed error information report.
[0094] The characteristic torsion rate and adjustment parameters are corrected based on the error information to obtain compensated characteristic torsion rate and error compensation parameters.
[0095] The characteristic torsion rate and adjustment parameters are corrected and compensated by using error information; specifically, an error compensation algorithm is used to adjust the characteristic torsion rate and adjustment parameters to reduce the influence of the error, and the compensated characteristic torsion rate and corresponding error compensation parameters are obtained.
[0096] For example, a Kalman filter or a particle filter algorithm may be used to process the error information, correct the characteristic torsion rate and adjust the parameters to obtain a compensation result with a smaller error.
[0097] In step S500, the compensated characteristic torsion and error compensation parameters are input into a preset dynamic analysis model, and a time series algorithm is executed in the dynamic analysis model to evaluate the time dependency of the compensated characteristic torsion, capture the pattern of sensor performance changes over time, and obtain a timing analysis result.
[0098] The time dependency is evaluated by inputting the compensated characteristic torsion rate and error compensation parameters into the dynamic analysis model. Specifically, a time series analysis algorithm, such as an ARIMA model or a long short-term memory network (LSTM), is used to analyze the compensated characteristic torsion rate, capture the pattern of sensor performance changes over time, and obtain the timing analysis results.
[0099] For example, an LSTM model can be used to input the compensated characteristic torsion rate and error compensation parameters into the model, and the trend and pattern of sensor performance changes over time can be obtained through training and prediction of time series data.
[0100] Based on the preset stability analysis algorithm, the stability analysis is performed on the timing analysis results to confirm the consistency and predictability of the characteristic torsion rate after compensation under different operating conditions, and obtain the stability analysis results.
[0101] By performing stability analysis on the timing analysis results, the stability of the characteristic torsion after compensation is evaluated; specifically, a preset stability analysis algorithm, such as Monte Carlo simulation or variance analysis, is used to evaluate the timing analysis results to confirm their consistency and predictability under different operating conditions, thereby obtaining stability analysis results.
[0102] For example, the Monte Carlo simulation method can be used to perform multiple random simulations on the timing analysis results to evaluate the fluctuation range and stability of the characteristic torsion rate after compensation under different conditions, and obtain a detailed stability analysis report.
[0103] The timing analysis results and stability analysis results are integrated by fusing the real-time feedback algorithm to obtain the performance analysis results that characterize the dynamic performance of the sensor.
[0104] By integrating the timing analysis results and the stability analysis results, the comprehensive performance analysis results of the sensor are obtained; specifically, a real-time feedback algorithm, such as incremental learning or adaptive filtering, is used to fuse the timing analysis results and the stability analysis results to generate a comprehensive performance analysis result that fully reflects the dynamic performance of the sensor.
[0105] For example, an incremental learning algorithm can be used to update and fuse the timing analysis results and the stability analysis results in real time to obtain a comprehensive performance analysis vector that can reflect the dynamic performance of the sensor under different conditions in real time.
[0106] Use the difference evaluation tool to evaluate the performance analysis results according to the comprehensive feature vector, establish the dynamic response law, and formulate calibration indicators based on the dynamic response law and preset market feedback data.
[0107] By utilizing differential evaluation tools, the performance analysis results are evaluated and the dynamic response law of the sensor is established. Specifically, differential evaluation methods, such as the differential method or relative error analysis, are used to evaluate the comprehensive characteristic vector and performance analysis results, determine the dynamic response law of the sensor under different conditions, and formulate calibration indicators in combination with market feedback data.
[0108] For example, the differential method can be used to perform differential analysis on the comprehensive feature vector and performance analysis results to identify the dynamic response characteristics of the sensor under different operating conditions. Then, based on market feedback data, a series of calibration indicators can be formulated to ensure that the performance of the sensor meets the actual application requirements.
[0109] By combining calibration indicators with error compensation parameters, a comprehensive calibration strategy for multi-dimensional force sensors is developed.
[0110] By combining calibration indicators and error compensation parameters, a comprehensive calibration strategy is formulated; specifically, the comprehensive calibration strategy includes adjusting the sensor's hardware parameters, software algorithms, and usage environment to ensure its high accuracy and high stability in practical applications.
[0111] For example, the sensor hardware can be adjusted according to the calibration indicators and error compensation parameters, such as recalibrating the sensor's zero point and full scale, and the software algorithm can be optimized, such as adjusting the filtering parameters and compensation algorithm. At the same time, the use environment can be adjusted, such as reducing environmental noise and temperature fluctuations, to ensure that the sensor can maintain high accuracy under various conditions.
[0112] In step S600, the geometric position information is analyzed using a fuzzy logic algorithm to obtain a spatial positioning error.
[0113] The geometric position information is analyzed and processed by using fuzzy logic algorithms; specifically, the fuzzy logic algorithm converts the geometric position information into a fuzzy set by fuzzifying the input data, and then calculates the spatial positioning error through the fuzzy reasoning process to reveal the positioning error of the sensor in the actual position.
[0114] For example, a fuzzy logic controller can be used to input the geometric position information into the controller. After fuzzification and fuzzy reasoning, a fuzzy set reflecting the spatial positioning error can be obtained. This set can intuitively reflect the positioning accuracy of the sensor at different positions.
[0115] The comprehensive calibration strategy is adjusted according to a preset piecewise linear correction algorithm to obtain calibration parameters.
[0116] The comprehensive calibration strategy is adjusted and optimized by using a preset piecewise linear correction algorithm; specifically, the piecewise linear correction algorithm reduces the complexity of the global correction by dividing the calibration data into several linear segments and independently correcting each linear segment, thereby obtaining more accurate calibration parameters.
[0117] For example, the sensor calibration data can be divided into several segments according to different force value ranges, and linear regression analysis can be performed on each segment of data to obtain the correction parameters of each segment. These parameters are then combined to form a global calibration parameter set for actual calibration.
[0118] The spatial positioning error and calibration parameters are input into a preset intelligent compensation model, and a genetic optimization algorithm is used in the intelligent compensation model to match the geometric position information and the calibration parameters to obtain matching analysis data and calibration instructions.
[0119] By inputting the spatial positioning error and calibration parameters into the intelligent compensation model, intelligent optimization and matching analysis are performed; specifically, the intelligent compensation model uses a genetic optimization algorithm to globally optimize the geometric position information and calibration parameters, find the optimal matching solution, and generate matching analysis data and calibration instructions.
[0120] For example, a genetic algorithm can be used to input spatial positioning errors and calibration parameters into the algorithm. After multiple iterations of optimization, an optimal matching solution can be found to generate corresponding calibration instructions and matching analysis data to ensure the calibration accuracy of the sensor.
[0121] The real-time calibration operation is performed according to the calibration instruction to obtain the real-time calibration data, and the real-time calibration data and the matching analysis data are subjected to terminal calibration analysis to obtain the calibration result.
[0122] Real-time calibration data is obtained by performing real-time calibration operations according to calibration instructions; specifically, the real-time calibration operation includes online adjustment and correction of the sensor, real-time recording of calibration data, and comparing these data with matching analysis data to obtain the final calibration result.
[0123] For example, in actual operation, the sensor can be adjusted according to the calibration instructions, the calibration data after each adjustment can be recorded, and these data can be compared with the matching analysis data to confirm the calibration effect, and finally the calibration result of the sensor can be obtained to ensure its accuracy in actual application.
[0124] In this embodiment, the real-time force data of the sensor is acquired through the data module, and the sensor is accurately positioned and calibrated using optical positioning technology to obtain its morphological parameters and spatial response characteristics; on this basis, the preprocessed data is processed using Kalman filtering and particle filtering algorithms to obtain accurate geometric position information; then, the nonlinear mapping algorithm is adjusted using geometric position constraints, the force data is reduced in dimension through kernel principal component analysis, and support vector machines are used to extract fusion force features and force feature correlations.
[0125] Through a series of feature association analysis and weighted processing, a low-dimensional force feature vector reflecting the force characteristics is obtained, and the geometric position constraints are integrated to generate a comprehensive feature vector. On this basis, the differential geometry analysis model is used to conduct an in-depth analysis of the comprehensive feature vector, extract the characteristic curvature and characteristic torsion, and determine the optimal characteristic curvature through the optimization algorithm; then, these adjusted features are input into the error compensation model, and the compensated data is obtained through back propagation network analysis, and the error source is identified and corrected. Next, the compensated features are evaluated in time series and analyzed in stability through the dynamic analysis model to obtain the dynamic performance results of the sensor.
[0126] Through real-time feedback algorithms and difference evaluation tools, comprehensive feature vectors and performance analysis results are evaluated to develop a comprehensive calibration strategy. Fuzzy logic algorithms and piecewise linear correction algorithms are used to calibrate the sensor in real time to ensure its high accuracy and stability under various operating conditions. This comprehensive calibration method not only improves the measurement accuracy of the multi-dimensional force sensor, but also enhances its adaptability in complex application environments.
[0127] Embodiment 3:
[0128] like Figure 2 As shown, the present application also discloses a calibration device 10 for a multi-dimensional force sensor, which includes an acquisition module 11, an extraction module 12, an adjustment module 13, a compensation module 14, an analysis module 15 and a matching module 16.
[0129] The acquisition module 11 is mainly used to acquire force data and geometric position information of the multi-dimensional force sensor.
[0130] The extraction module 12 is mainly used to extract features from the force data using a preset nonlinear mapping algorithm to obtain a comprehensive feature vector.
[0131] The adjustment module 13 is mainly used to perform differential geometry analysis on the comprehensive characteristic vector to obtain characteristic curvature and characteristic torsion, and adjust the characteristic curvature based on the optimization algorithm to obtain the adjusted characteristic curvature and adjustment parameters.
[0132] The compensation module 14 is mainly used to perform error compensation on the characteristic torsion based on the adjusted characteristic curvature and the adjustment parameter to obtain the compensated characteristic torsion and the error compensation parameter.
[0133] The analysis module 15 is mainly used to dynamically analyze and process the compensated characteristic torsion and error compensation parameters to obtain analysis results; the analysis results are processed according to the comprehensive characteristic vector to obtain calibration indicators, and a comprehensive calibration strategy is formulated based on the calibration indicators and adjustment parameters.
[0134] The matching module 16 is mainly used to perform matching analysis on the geometric position information and the comprehensive calibration strategy to obtain the calibration result.
[0135] In this embodiment, by combining the acquisition module 11, the extraction module 12, the adjustment module 13, the compensation module 14, the analysis module 15 and the matching module 16, the high-precision calibration of the multi-dimensional force sensor is realized. Specifically, the force data and geometric position information of the multi-dimensional force sensor are acquired in real time by the acquisition module 11, and then the acquired force data is processed by the nonlinear mapping algorithm of the extraction module 12 to extract the comprehensive feature vector. Then, the adjustment module 13 performs differential geometry analysis on the comprehensive feature vector to obtain the characteristic curvature and characteristic torsion, and adjusts the characteristic curvature by the optimization algorithm to generate the adjusted characteristic curvature and the corresponding adjustment parameters. The compensation module 14 performs error compensation based on these adjusted parameters to obtain the compensated characteristic torsion and error compensation parameters. Subsequently, the analysis module 15 performs dynamic analysis and processing on the compensated characteristic torsion and error compensation parameters to obtain the analysis results, and processes these results according to the comprehensive feature vector, formulates calibration indicators, and forms a comprehensive calibration strategy. Finally, the matching module 16 matches and analyzes the geometric position information with the comprehensive calibration strategy to generate the final calibration result. Through this systematic calibration device, the calibration accuracy and adaptability of the multi-dimensional force sensor can be significantly improved, ensuring its stable and reliable performance in practical applications.
[0136] It should be noted that technicians in the relevant technical field can clearly understand that for the convenience and simplicity of description, the specific working process of the above-described device and each module can refer to the corresponding process in the aforementioned multi-dimensional force sensor calibration method embodiment, and will not be repeated here.
[0137] Embodiment 4:
[0138] like Figure 3 As shown, the present application also discloses an electronic device 20, including a memory 21 and a processor 22, wherein the memory 21 stores a computer program that can be run on the processor 22, and when the processor 22 executes the computer program, the calibration method of the multi-dimensional force sensor of Example 1 is implemented.
[0139] In this embodiment, the automatic calibration of the multi-dimensional force sensor is realized by integrating the memory 21 and the processor 22 in the electronic device 20, and storing the computer program that can be run on the processor 22 in the memory 21. Specifically, when the processor 22 executes the computer program, the multi-dimensional force sensor will be systematically calibrated and calibrated according to the method described in Example 1. First, the processor 22 obtains real-time force data and geometric position information from the sensor, and then uses a preset algorithm to extract features from the force data to obtain a comprehensive feature vector. Then, the processor 22 performs differential geometry analysis on these feature vectors to obtain characteristic curvature and characteristic torsion, and adjusts the characteristic curvature through an optimization algorithm to generate adjusted characteristic curvature and parameters. Subsequently, the processor 22 performs error compensation and dynamic analysis, and finally obtains the calibration results and calibration strategy. In this way, the electronic device 20 can efficiently and accurately complete the calibration task of the multi-dimensional force sensor, ensuring its accuracy and stability in actual use.
[0140] Embodiment 5:
[0141] The present application also discloses a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the processor executes the calibration method of the multi-dimensional force sensor as described in Example 1.
[0142] In this embodiment, by storing a computer program on a computer-readable storage medium, when the processor runs the program, the calibration method of the multi-dimensional force sensor as described in Example 1 is executed. Specifically, the computer program instructs the processor to obtain real-time force data and geometric position information from the sensor according to preset steps, and then performs nonlinear mapping and feature extraction to obtain a comprehensive feature vector. The processor further performs differential geometry analysis and optimization processing on these feature vectors to generate adjusted characteristic curvatures and adjustment parameters. Subsequently, the program instructs the processor to perform error compensation, analyze and process the compensated characteristic torsion and parameters, and finally generate calibration results and calibration strategies. Through the application of this computer-readable storage medium, efficient calibration of multi-dimensional force sensors can be easily implemented on different devices to ensure the measurement accuracy and reliability of the sensor.
[0143] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A calibration method for a multi-dimensional force sensor, characterized in that: include: Obtain force data and geometric position information of multi-dimensional force sensors; Using a preset nonlinear mapping algorithm to extract features from the force data to obtain a comprehensive feature vector; Performing differential geometry analysis on the comprehensive characteristic vector to obtain characteristic curvature and characteristic torsion, and adjusting the characteristic curvature based on a preset optimization algorithm to obtain an adjusted characteristic curvature and adjustment parameters; Performing error compensation on the characteristic torsion based on the adjusted characteristic curvature and the adjustment parameter to obtain a compensated characteristic torsion and an error compensation parameter; Performing dynamic analysis on the compensated characteristic torsion and the error compensation parameter to obtain an analysis result; Processing the analysis result according to the comprehensive feature vector to obtain a calibration index, and formulating a comprehensive calibration strategy based on the calibration index and the adjustment parameter; The geometric position information and the comprehensive calibration strategy are matched and analyzed to obtain a calibration result.
2. The calibration method of the multi-dimensional force sensor according to claim 1, characterized in that: The step of obtaining the force data and geometric position information of the multi-dimensional force sensor includes: Obtain real-time force data from multi-dimensional force sensors based on the data module; The multi-dimensional force sensor is positioned based on optical positioning technology, and the geometric sensitive area of the multi-dimensional force sensor is calibrated to obtain the morphological parameters and spatial response characteristics of the multi-dimensional force sensor; The morphological parameters and the spatial response characteristics are integrated to obtain pre-processed data, and the pre-processed data are calculated using a Kalman filter algorithm and a particle filter algorithm to obtain geometric position information.
3. The calibration method of the multi-dimensional force sensor according to claim 1, characterized in that: The step of extracting features from the force data using a preset nonlinear mapping algorithm to obtain a comprehensive feature vector includes: The geometric position information is used as a geometric position constraint, and a preset nonlinear mapping algorithm is adjusted based on the geometric position constraint, the force data is input into the adjusted nonlinear mapping algorithm, and a kernel principal component analysis is used in the adjusted nonlinear mapping algorithm to perform a dimensionality reduction analysis on the force data to obtain a dimensionality reduction result; Using a support vector machine to extract features from the dimension reduction results, and obtaining fusion force features and force feature correlations; Performing feature correlation analysis on the fusion force feature and the force feature correlation to obtain a correlation feature vector; Performing position-sensitive weighting on the associated feature vector through the geometric position constraint to obtain a position-weighted force feature vector; Inputting the position-weighted force feature vector into a preset feature mapping network to obtain a low-dimensional force feature vector reflecting force characteristics; The low-dimensional force feature vector is integrated with the geometric position constraint to generate a comprehensive feature vector representing the multi-dimensional force perception capability.
4. The calibration method of the multi-dimensional force sensor according to claim 1, characterized in that: The step of performing differential geometry analysis on the comprehensive characteristic vector to obtain characteristic curvature and characteristic torsion, and adjusting the characteristic curvature based on a preset optimization algorithm to obtain the adjusted characteristic curvature and adjustment parameters includes: The comprehensive feature vector is input into a preset differential geometry analysis model, and the comprehensive feature vector is locally analyzed by surface theory in the differential geometry analysis model to extract characteristic curvature; wherein the characteristic curvature reflects the bending degree of the multi-dimensional force sensor under different forces; The characteristic torsion is calculated by using the torsion theory and combining the characteristic curvature; wherein the characteristic torsion reveals the intrinsic properties of the geometric deformation of the force sensor; The optimal characteristic curvature of the characteristic torsion characterizing the geometric deformation of the sensor is determined by an optimization algorithm, the optimal characteristic curvature is used as the adjusted characteristic curvature, and the adjustment parameters of the optimal characteristic curvature of the characteristic torsion characterizing the geometric deformation of the sensor are determined and recorded.
5. The calibration method of the multi-dimensional force sensor according to claim 1, characterized in that: The step of performing error compensation on the characteristic torsion based on the adjusted characteristic curvature and the adjustment parameter to obtain the compensated characteristic torsion and the error compensation parameter comprises: Inputting the adjusted characteristic curvature and the adjustment parameter into an error compensation model, and analyzing the adjusted characteristic curvature and the adjustment parameter by using a back propagation network in the error compensation model to obtain result data; Extracting the error source that affects the accuracy of the characteristic torsion from the result data to obtain error information; wherein the error source includes any one of the following: system error, environmental error, and measurement error; The characteristic torsion rate and adjustment parameters are corrected based on the error information to obtain compensated characteristic torsion rate and error compensation parameters.
6. The calibration method of the multi-dimensional force sensor according to claim 1, characterized in that: The compensated characteristic torsion and the error compensation parameter are dynamically analyzed to obtain an analysis result; The step of processing the analysis result according to the comprehensive feature vector to obtain a calibration index, and formulating a comprehensive calibration strategy based on the calibration index and the adjustment parameter includes: Inputting the compensated characteristic torsion rate and the error compensation parameter into a preset dynamic analysis model, executing a time series algorithm in the dynamic analysis model to evaluate the time dependency of the compensated characteristic torsion rate, capturing the pattern of sensor performance changing over time, and obtaining a time series analysis result; Performing stability analysis on the timing analysis result based on a preset stability analysis algorithm to confirm the consistency and predictability of the characteristic torsion rate after compensation under different operating conditions, and obtaining a stability analysis result; The timing analysis result and the stability analysis result are integrated by fusing a real-time feedback algorithm to obtain a performance analysis result characterizing the dynamic performance of the sensor; Using a difference evaluation tool to evaluate the performance analysis results according to a comprehensive feature vector, establishing a dynamic response law, and formulating calibration indicators based on the dynamic response law and preset market feedback data; By combining the calibration index with the error compensation parameter, a comprehensive calibration strategy for the multi-dimensional force sensor is developed.
7. The calibration method of a multi-dimensional force sensor according to claim 1, characterized in that: The step of performing matching analysis on the geometric position information and the comprehensive calibration strategy to obtain a calibration result includes: Analyzing the geometric position information using a fuzzy logic algorithm to obtain a spatial positioning error; The comprehensive calibration strategy is adjusted according to a preset piecewise linear correction algorithm to obtain calibration parameters; Inputting the spatial positioning error and the calibration parameter into a preset intelligent compensation model, and matching the geometric position information and the calibration parameter using a genetic optimization algorithm in the intelligent compensation model to obtain matching analysis data and calibration instructions; A real-time calibration operation is performed according to the calibration instruction to obtain real-time calibration data, and terminal calibration analysis is performed on the real-time calibration data and the matching analysis data to obtain a calibration result.
8. A calibration device for a multi-dimensional force sensor, characterized in that: include: An acquisition module is used to obtain force data and geometric position information of a multi-dimensional force sensor; An extraction module, used for performing feature extraction on the force data using a preset nonlinear mapping algorithm to obtain a comprehensive feature vector; An adjustment module, used for performing differential geometry analysis on the comprehensive characteristic vector to obtain characteristic curvature and characteristic torsion, and adjusting the characteristic curvature based on an optimization algorithm to obtain an adjusted characteristic curvature and adjustment parameters; A compensation module, used for performing error compensation on the characteristic torsion based on the adjusted characteristic curvature and the adjustment parameter to obtain a compensated characteristic torsion and an error compensation parameter; An analysis module, used for dynamically analyzing the compensated characteristic torsion and the error compensation parameter to obtain an analysis result; Processing the analysis result according to the comprehensive feature vector to obtain a calibration index, and formulating a comprehensive calibration strategy based on the calibration index and the adjustment parameter; The matching module is used to perform matching analysis on the geometric position information and the comprehensive calibration strategy to obtain a calibration result.
9. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, the calibration method of the multi-dimensional force sensor according to any one of claims 1 to 7 is implemented.
10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the processor executes the calibration method of the multi-dimensional force sensor as claimed in any one of claims 1 to 7.
